English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Deep learning based video frame interpolation (VIF) method, aiming tosynthesis the intermediate frames to enhance video quality, have been highlydeveloped in the past few years. This paper investigates the adversarialrobustness of VIF models. We apply adversarial attacks to VIF models and findthat the VIF models are very vulnerable to adversarial examples. To improveattack efficiency, we suggest to make full use of the property of video frameinterpolation task. The intuition is that the gap between adjacent frames wouldbe small, leading to the corresponding adversarial perturbations being similaras well. Then we propose a novel attack method named Inter-frame AccelerateAttack (IAA) that initializes the perturbation as the perturbation for theprevious adjacent frame and reduces the number of attack iterations. It isshown that our method can improve attack efficiency greatly while achievingcomparable attack performance with traditional methods. Besides, we also extendour method to video recognition models which are higher level vision tasks andachieves great attack efficiency.

Inter-frame Accelerate Attack against Video Interpolation Models